Research ArticleImmunology

Noncanonical STAT1 phosphorylation expands its transcriptional activity into promoting LPS-induced IL-6 and IL-12p40 production

See allHide authors and affiliations

Science Signaling  24 Mar 2020:
Vol. 13, Issue 624, eaay0574
DOI: 10.1126/scisignal.aay0574

Inflammatory internalization

When bound to its ligand LPS, the cell surface pattern recognition receptor TLR4 stimulates the production of proinflammatory cytokines. However, endocytosed TLR4 signals through different effectors to drive production of interferon-β (IFN-β), an antiviral cytokine. Metwally et al. found an alternative signaling pathway activated by endocytosed TLR4 in mouse and human macrophages that contributed to the production of the proinflammatory cytokines IL-6 and IL-12p40 independently of IFN-β. Endocytosis of TLR4 led to the noncanonical phosphorylation of the transcriptional regulator STAT1, which altered its target DNA motif. STAT1 stimulated the expression of the genes encoding IL-12p40 and the protein ARID5A, which increased the stability of IL6 mRNA. Together, these findings suggest how endocytosed TLR4 contributes to proinflammatory cytokine production, which may have implications for the use of vaccine adjuvants that target this receptor.

Abstract

The lipopolysaccharide (LPS)–induced endocytosis of Toll-like receptor 4 (TLR4) is an essential step in the production of interferon-β (IFN-β), which activates the transcription of antiviral response genes by STAT1 phosphorylated at Tyr701. Here, we showed that STAT1 regulated proinflammatory cytokine production downstream of TLR4 endocytosis independently of IFN-β signaling and the key proinflammatory regulator NF-κB. In human macrophages, TLR4 endocytosis activated a noncanonical phosphorylation of STAT1 at Thr749, which subsequently promoted the production of interleukin-6 (IL-6) and IL-12p40 through distinct mechanisms. STAT1 phosphorylated at Thr749 activated the expression of the gene encoding ARID5A, which stabilizes IL6 mRNA. Moreover, STAT1 phosphorylated at Thr749 directly enhanced transcription of the gene encoding IL-12p40 (IL12B). Instead of affecting STAT1 nuclear translocation, phosphorylation of Thr749 facilitated the binding of STAT1 to a noncanonical DNA motif (5′-TTTGANNC-3′) in the promoter regions of ARID5A and IL12B. The endocytosis of TLR4 induced the formation of a complex between the kinases TBK1 and IKKβ, which mediated the phosphorylation of STAT1 at Thr749. Our data suggest that noncanonical phosphorylation in response to LPS confers STAT1 with distinct DNA binding and gene-regulatory properties that promote both IL12B expression and IL6 mRNA stabilization. Thus, our study provides a potential mechanism for how TLR4 endocytosis might regulate proinflammatory cytokine production.

INTRODUCTION

Toll-like receptors (TLRs) comprise a family of pattern recognition receptors (PRRs), which play a crucial role in the initiation of the innate immune response (1). The subcellular localization of activated TLRs is essential for the engagement of specific adaptor molecules that initiate signaling cascades and culminate in an appropriate transcriptional response (2). This paradigm is evident in the case of TLR4, a mammalian receptor for bacterial lipopolysaccharide (LPS). TLR4 exploits different adaptors to induce distinct signaling pathways, thus expanding the repertoire of transcribed genes and potentiating the immune response (3). At the plasma membrane, LPS triggers TLR4 dimerization and recruits the adaptor protein MYD88 (myeloid differentiation marker 88), which activates proinflammatory signaling pathways, such as those mediated by the transcription factor nuclear factor κB (NF-κB) and members of the mitogen-activated protein kinase (MAPK) family (4). On the other hand, the association of activated TLR4 with co-receptors such as CD14 promotes its endocytosis, which is clathrin and dynamin dependent (5, 6). In the endosomal compartment, TLR4 promotes signaling mediated by the adaptor protein TRIF [TIR domain–containing adaptor-inducing interferon-β (IFN-β)], which activates the transcriptional regulator interferon regulatory factor 3 (IRF3) and the subsequent expression of genes encoding type I IFNs (7).

Binding of type I IFNs to their receptor [IFN-α/β receptor (IFNAR)] activates Janus kinase 1 (JAK1) and tyrosine kinase 2 (Tyk2), which, in turn, promotes Tyr701 phosphorylation of signal transducer and activator 1 (STAT1), a key step for its transcriptional activity (8). This phosphorylation event results in the formation of STAT1 homodimers, which bind to GAS (IFN-γ–activated sequence) sites, or the binding of STAT1 with STAT2 and IRF9 to form the IFN-stimulated gene factor 3 (ISGF3) complex, which binds to ISRE (IFN-stimulated response element) sites, which induces the transcription of multiple IFN-stimulated genes (ISGs) (9). Although type I IFNs play a central role in the antiviral immune response, their contribution to host defense against bacterial pathogens is elusive, with mounting evidence indicating a detrimental effect (10). Many reports have shown that type I IFN signaling is dispensable for the production of proinflammatory cytokines (1113). TRIF signaling promotes proinflammatory cytokine production through sustaining a late-phase activation of NF-κB (14, 15). However, this idea is challenged by several observations showing that neither TRIF deficiency nor interfering with TLR4 endocytosis affects the kinetics of NF-κB activation (1618).

Previous reports showed that the antipsychotic drug chlorpromazine prevents LPS-induced lethality in murine models (19). However, chlorpromazine does not inhibit LPS-induced production of tumor necrosis factor–α (TNF-α) or interleukin-1β (IL-1β) by the peripheral blood mononuclear cells (PBMCs) of healthy volunteers (20). Later studies demonstrated that chlorpromazine is a potent inhibitor of clathrin-mediated endocytosis (21, 22). Our group previously showed that chlorpromazine inhibits the production of IL-6, but not TNF-α, by murine macrophages. Chlorpromazine reduces expression of the gene encoding AT-rich interactive domain-containing protein 5a (Arid5a), which selectively stabilizes the 3′ untranslated region (3′UTR) of IL6 mRNA (23). Consistently, Tatematsu et al. (16) showed that disruption of TLR4 endocytosis diminishes IFN-β and IL-6 production, but not NF-κB activation or TNF-α production in macrophages. In addition, loss of Arid5a, similar to that of TRIF, confers enhanced survival and diminished proinflammatory cytokine production in LPS-induced sepsis (24). We reasoned that TLR4 endosomal signaling might regulate proinflammatory cytokine production through promoting ARID5A expression independently of NF-κB activation. Here, we found that TLR4 endocytosis enhanced the production of IL-6 and IL-12p40 independently of NF-κB activation through stimulating a noncanonical phosphorylation of STAT1 at Thr749. STAT1-pThr749 enhanced the stabilization of IL6 mRNA through activating transcription of the gene encoding ARID5A. On the other hand, STAT1-pThr749 directly augmented the transcription of IL12B through binding to its promoter region.

RESULTS

LPS-induced TLR4 endocytosis promotes ARID5A, IL-6, and IL-12p40 production

Kagan et al. (7) showed that interfering with dynamin guanosine triphosphatases (GTPases) disrupted TLR4 endocytosis and the subsequent TRAM-TRIF–mediated production of IFN-β. We therefore tested the effect of two different dynamin inhibitors, dynasore, which targets the GTPase domain, and MitMAB, which targets the pleckstrin homology (PH) domain of dynamin (25), on LPS-induced proinflammatory cytokine production in human macrophages (Fig. 1 and fig. S1). Both inhibitors disrupted TLR4 endocytosis and the subsequent expression of the gene encoding IFN-β, with MitMAB showing a more potent inhibitory effect on IFNB expression (fig. S1, A and B). We next examined the effect of MitMAB on ARID5A expression in response to LPS. MitMAB treatment resulted in reduced ARID5A expression and ARID5A protein abundance in human monocyte-derived macrophages (MDMs) (Fig. 1, A and B) and differentiated THP-1 (dTHP-1) cells (fig. S1, C and F). MitMAB treatment caused three distinct patterns of cytokine-encoding gene expression upon LPS stimulation. First, some genes were not affected by MitMAB, such as TNF and IL1B. Second, the expression of some cytokine-encoding genes was abolished by MitMAB, such as IFNB and CCL5. Third, the expression of some cytokine-encoding genes was inhibited to a lesser extent by MitMAB, including IL6 and IL12B (Fig. 1C and fig. S1D). TLR4 endosomal signaling is mediated by TRIF, rather than MYD88 (5, 7). Therefore, we hypothesized that IFNB and CCL5 expression exclusively depended on TRIF, whereas TNF and IL1B expression exclusively depended on MYD88 signaling. On the other hand, both TRIF and MYD88 signaling might culminate in the expression of IL6 and IL12B upon LPS stimulation. To further investigate this hypothesis, we evaluated the expression of different cytokine-encoding genes upon exposure to LPS, which promotes both MYD88 and TRIF signaling, or P3C (Pam3CSK4; a TLR2 ligand), which promotes MYD88, but not TRIF signaling (1). Consistent with our hypothesis, we observed that LPS induced TNF and IL1B expression similarly to P3C (Fig. 1D). In contrast, LPS, but not P3C, induced the expression of IFNB and CCL5. Note that LPS increased the amounts of IL-6 and IL-12p40 mRNA and protein to a greater extent than did P3C (Fig. 1, D and E). Moreover, LPS, but not P3C, induced ARID5A expression in MDMs (Fig. 1F) and dTHP-1 cells (fig. S1E). Furthermore, MitMAB inhibited ARID5A protein production in MDMs (Fig. 1G) and dTHP-1 cells (fig. S1F) treated with LPS; P3C had no effect on ARID5A protein abundance.

Fig. 1 LPS-induced TLR4 endocytosis promotes ARID5A, IL-6, and IL-12p40 production.

(A to C) MDMs were stimulated with LPS (100 ng/ml) in the presence or absence of 10 μM MitMAB. (A and C) Total RNA was isolated at the indicated times. The indicated transcripts (mRNAs) were quantified by quantitative real-time polymerase chain reaction (qRT-PCR) analysis. Data are representative of three independent experiments (six donors per experiment) and are presented as means ± SD. (B) Whole-cell lysates were harvested 3 hours after stimulation and were separated by SDS–polyacrylamide gel electrophoresis (PAGE). The indicated endogenous proteins were detected by Western blotting analysis. Blots are representative of three independent experiments. (D to F) MDMs were left unstimulated or were stimulated with LPS (100 ng/ml) or P3C (100 ng/ml). (D and F) Total RNA was isolated at the indicated times. The indicated transcripts were quantified by qRT-PCR analysis. Data are representative of three independent experiments (six donors per experiment) and are presented as means ± SD. (E) Cell-free supernatants were harvested 24 hours after stimulation and were analyzed by enzyme-linked immunosorbent assay (ELISA) for the indicated cytokines. Data are representative of three independent experiments (six donors per experiment) and are presented as means ± SD. (G) MDMs were left unstimulated or were stimulated with LPS (100 ng/ml) or P3C (100 ng/ml) in the presence or absence of 10 μM MitMAB. Whole-cell lysates were harvested 3 hours after stimulation and separated by SDS-PAGE. The indicated endogenous proteins were detected by Western blotting analysis. Blots are representative of three independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 as measured by one-way analysis of variance (ANOVA) with post hoc Tukey’s test (A and C to F). n.s, not significant.

We next examined whether TLR4 endocytosis in MDMs promoted IL-6 and IL-12p40 expression through ARID5A. We found that small interfering RNA (siRNA)–mediated knockdown of ARID5A in LPS-stimulated MDMs statistically significantly attenuated IL6 expression, but not that of IL12B or TNF (fig. S2A). In addition, ARID5A selectively stabilized IL6 mRNA, but not IL12B or TNF mRNAs (fig. S2, B and C). We next performed deletion analysis of the 3′UTR of IL6 to identify the region to which ARID5A bound. We found that ARID5A exerted its stabilizing activity through binding to the IL6 3′UTR at an AU-rich element that is conserved between the murine and human genes (fig. S2, D to F). Consistently, we found in our pull-down assays that ARID5A bound to biotinylated nucleotides (bio-oligos) containing this same identified region (fig. S2G). Together, our results indicate that LPS-induced TLR4 endocytosis promotes IL-12p40 and IL-6 production and that ARID5A stabilizes the mRNA of the latter.

TLR4 endocytosis–activated IFN-β–JAK–STAT1-pTyr701 signaling is dispensable for ARID5A expression

We next sought to identify the signaling pathway governing ARID5A expression. First, we examined whether ARID5A expression required MYD88 or TRIF. Consistent with the dependency of ARID5A expression on TLR4 endocytosis, we found that MYD88−/ dTHP-1 cells retained their ability to express ARID5A mRNA and protein upon LPS stimulation (Fig. 2, A and B). In contrast, the siRNA-mediated knockdown of TRIF in LPS-stimulated dTHP-1 cells abolished ARID5A expression (Fig. 2, C and D). Stimulation of TLR4, but not TLR2, induces IFN-β production and the phosphorylation of STAT1 at Tyr701 (26). Consistently, we found that LPS, but not P3C, induced Tyr701 phosphorylation in STAT1 in dTHP-1 cells. In addition, MitMAB abolished the generation of STAT1-pTyr701 in LPS-stimulated dTHP-1 cells (fig. S3A). Moreover, MYD88−/− dTHP-1 cells expressed IFNB transcripts and exhibited STAT1-pTyr701 upon LPS stimulation similar to wild-type (WT) cells (fig. S3, B and C). The similar expression patterns of ARID5A and IFNB, coupled with their dependency on TLR4 endocytosis–mediated TRIF signaling, prompted us to examine whether the IFN-β–STAT1-pTyr701 axis promoted ARID5A expression. We found that siRNA-mediated knockdown of STAT1 inhibited ARID5A expression in LPS-stimulated macrophages (Fig. 2, E and F), suggesting that STAT1-pTyr701 might induce ARID5A expression.

Fig. 2 TLR4 endocytosis–dependent IFN-β–JAK–STAT1-pTyr701 signaling is dispensable for ARID5A expression.

(A and B) WT or MYD88−/− dTHP-1 cells were left unstimulated or were stimulated with LPS (100 ng/ml) for 3 hours. (A) Total RNA was isolated and ARID5A transcripts were quantified by qRT-PCR analysis. Data are representative of three independent experiments and are presented as means ± SD. (B) Whole-cell lysates were harvested and separated by SDS-PAGE. The indicated endogenous proteins were detected by Western blotting analysis. Blots are representative of three independent experiments. (C to F) WT dTHP-1 cells were transfected with scrambled siRNA or siRNAs targeting TRIF (C and D) or STAT1 (E and F). Forty-eight hours later, the cells were then left unstimulated or were stimulated with LPS (100 ng/ml). (C and E) Whole-cell lysates from the indicated transfected cells were separated by SDS-PAGE and analyzed by Western blotting. Blots are representative of three independent experiments. (D and F) Total RNA was isolated from the indicated cells 3 hours after stimulation. ARID5A transcripts were quantified by qRT-PCR analysis. Data are representative of three independent experiments and are presented as means ± SD. (G and H) WT or IFNAR2−/− dTHP-1 cells were left unstimulated or were stimulated with LPS (100 ng/ml) for 3 hours. (G) Total RNA was isolated and ARID5A transcripts were quantified by qRT-PCR analysis. Data are representative of three independent experiments and are presented as means ± SD. (H) Whole-cell lysates were harvested and separated by SDS-PAGE. The indicated endogenous proteins were detected by Western blotting analysis. Blots are representative of three independent experiments. (I) IFNAR2−/− dTHP-1 cells were transfected with scrambled siRNA or siRNA targeting STAT1. Forty-eight hours later, the cells were then left unstimulated or were stimulated with LPS (100 ng/ml). Whole-cell lysates were harvested 3 hours after stimulation and separated by SDS-PAGE. The indicated endogenous proteins were then detected by Western blotting analysis. Blots are representative of three independent experiments. (J and K) MDMs were stimulated with the indicated concentrations of human IFN-β for 1 hour (J) or were stimulated with human IFN-β (25 ng/ml) for the indicated times (K). Whole-cell lysates were harvested and separated by SDS-PAGE. The indicated endogenous proteins were detected by Western blotting analysis. Blots are representative of three independent experiments. (L and M) WT dTHP-1 cells were transfected with scrambled siRNA or siRNA targeting IRF3. Forty-eight hours later, the cells were then left unstimulated or were stimulated with LPS (100 ng/ml). (L) Whole-cell lysates from the transfected cells were separated by SDS-PAGE and analyzed by Western blotting. Blots are representative of three independent experiments. (M) Total RNA was isolated, and the indicated transcripts were quantified by qRT-PCR analysis. Data are representative of three independent experiments and are presented as means ± SD. (N) MDMs were transfected with scrambled siRNA or siRNA targeting IRF3. Forty-eight hours later, the cells were then left unstimulated or were stimulated with LPS (100 ng/ml) for 3 hours. Whole-cell lysates were harvested and separated by SDS-PAGE. The indicated endogenous proteins were detected by Western blotting. Blots are representative of three independent experiments. **P < 0.01, ***P < 0.001, ****P < 0.0001 as measured by one-way ANOVA with post hoc Tukey’s test (A, D, F, G, and M).

To further investigate this hypothesis, we used IFNAR2−/− THP-1 cells, which were validated not to generate STAT1-pTyr701 upon LPS stimulation (fig. S3D). Previous studies show that IFN-β–STAT1-pTyr701 signaling promotes IL6 expression (2729). Consistently, we found that IFNAR2−/− THP-1 cells had less IL6 mRNA upon LPS stimulation than did LPS-treated WT cells (fig. S3E). Unexpectedly, LPS-stimulated IFNAR2−/− dTHP-1 cells expressed ARID5A mRNA and protein in similar amounts to those in WT cells (Fig. 2, G and H). Cheon and Stark (30) previously showed that STAT1 induces immune regulatory gene expression independently of Tyr701 phosphorylation. We found that siRNA-mediated knockdown of STAT1 abolished ARID5A expression in LPS-stimulated IFNAR2−/− dTHP-1 cells (Fig. 2I), indicating that STAT1, but not STAT1-pTyr701, promoted ARID5A expression. Moreover, siRNA-mediated knockdown of STAT1 in IFNAR2−/− dTHP-1 cells statistically significantly attenuated LPS-stimulated IL6 expression (fig. S3F), suggesting that STAT1 promotes IL6 expression through another mechanism independently of STAT1-pTyr701. Furthermore, IFN-β–stimulated human MDMs generated STAT1-pTyr701, whereas they did not express ARID5A protein (Fig. 2, J and K). In addition, tofacitinib (a chemical inhibitor of JAK) abolished STAT1-pTyr701 generation in LPS-stimulated dTHP-1 cells, whereas it did not affect the LPS-induced increase in ARID5A protein abundance (fig. S3G). These findings suggest that LPS-induced IFN-β–STAT1-pTyr701 signaling is unlikely to promote ARID5A expression. We next examined whether IRF3 stimulated ARID5A expression. We found that siRNA-mediated knockdown of IRF3 in LPS-stimulated macrophages reduced the expression of IFNB and CCL5 and the generation of STAT1-pTyr701, whereas ARID5A mRNA and protein abundances were unperturbed (Fig. 2, L to N). In addition, IRF3 did not show any transcriptional activity on the ARID5A promoter region (fig. S3H). Together, our data indicate that TLR4-TRIF signaling regulates ARID5A expression in a STAT1-dependent manner, but not through the canonical pathway of IRF3 activation by IFN-β–STAT1-pTyr701 signaling.

Noncanonical phosphorylation of STAT1 at Thr749 activates ARID5A transcription

To determine how STAT1 regulated ARID5A expression, we generated a STAT1 knockdown THP-1 cell line using lentivirus-based, STAT1-specific short hairpin RNAs (shRNAs). We observed a specific reduction in STAT1 protein abundance in these cells (Fig. 3A). Functionally, we validated the effect of STAT1 deficiency in the THP-1 cells, because they showed reduced expression of antiviral response genes, such as IRF1, RSAD2, and IFIT2, in response to LPS (Fig. 3B). We observed that STAT1 knockdown dTHP-1 cells exhibited reduced amounts of ARID5A mRNA and protein upon LPS stimulation (Fig. 3, C and D). Note that STAT1 knockdown did not affect NF-κB or IRF3 signaling, as LPS-induced IκBα degradation and the phosphorylation and nuclear translocation of NF-κB and IRF3 were unaffected (Fig. 3, E and F).

Fig. 3 Noncanonical phosphorylation of STAT1 at Thr749 activates ARID5A transcription.

(A) Whole-cell lysates from WT dTHP-1 cells transduced with lentiviruses expressing scrambled shRNA or shRNA targeting STAT1 were separated by SDS-PAGE, and the indicated endogenous proteins were analyzed by Western blotting. Blots are representative of three independent experiments. (B to F) dTHP-1 cells expressing the indicated shRNAs were left unstimulated or were stimulated with LPS (100 ng/ml) for 3 hours (B to D) or 1 hour (E and F). (B and C) Total RNA was isolated, and the indicated transcripts were quantified by qRT-PCR analysis. Data are representative of three independent experiments and are presented as means ± SD. (D and F) Whole-cell lysates were separated by SDS-PAGE, and the indicated endogenous proteins were detected by Western blotting. (E) Cellular cytoplasmic and nuclear fractions were isolated and analyzed by Western blotting. Blots in (D) to (F) are representative of three independent experiments. (G to J) Measurement of the luciferase activity of U3A cells 48 hours after transfection with a luciferase reporter plasmid containing the human ARID5A promoter, together with control plasmid (EV) or an expression plasmid for WT STAT1 or one of its mutants. Results are presented relative to Renilla luciferase activity. Data are representative of three independent experiments and are presented as means ± SD. (K) Homology model of human STAT1 protein. The protein chain is colorized with a rainbow spectrum from the N-terminal (blue) to C-terminal (red) regions. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 as measured by one-way ANOVA with post hoc Tukey’s test (B, C, and H to J) or by paired Student’s t test (G).

These observations suggested that STAT1 might directly activate ARID5A expression. To further investigate this hypothesis, we performed reporter assays and found that STAT1 exerted transcriptional activity on the ARID5A promoter region (Fig. 3G). Phosphorylation of the Tyr701 or Ser727 residues in the C-terminal transactivation domain (TAD) of STAT1 is critical for its transcriptional activity (31). We next examined the transcriptional activity of STAT1 and its Tyr→Ala and Ser→Ala (Y/A and S/A) nonphospho-mimetic mutants Y701A, S727A, or both [Y701A/S727A, double mutant (DoMut)] on the ARID5A promoter region. Consistent with our findings that STAT1-pTyr701 was dispensable for ARID5A expression, we observed that the Y701A, S727A, and DoMut STAT1 variants showed transcriptional activities on the ARID5A promoter that were comparable to that of WT STAT1, whereas TAD-deficient STAT1 (NoTAD) failed to induce ARID5A promoter activity (Fig. 3H).

We next sought to identify how STAT1 promoted ARID5A expression. Cheon and Stark (30) showed that unphosphorylated STAT1 prolongs the expression of subsets of IFN-induced immune regulatory genes. However, it is unlikely that unphosphorylated STAT1 induces ARID5A expression in LPS-stimulated macrophages for two reasons. First, unphosphorylated STAT1 activity is a result of IFN signaling. However, ARID5A expression did not require IFN signaling (Fig. 2, G to N). Second, the effect of unphosphorylated STAT1 appears in 48 to 72 hours after IFN stimulation. The duration of our experiments is not sufficiently long to enable STAT1 accumulation, and maximal ARID5A expression was observed at 3 to 6 hours (Fig. 1F).

We next sought to identify whether STAT1 TAD phosphorylation sites other than the canonical Tyr701 and Ser727 residues were responsible for ARID5A promoter transcriptional activity. tenOever et al. (32) showed that the phosphorylation of Ser708 in STAT1 by IKKɛ is required for ISGF3 formation and the transcriptional activity of STAT1 on a distinct subset of antiviral genes. We next generated a series of STAT1 Ser→Ala and Thr→Ala (S/A and T/A) nonphospho-mimetic mutants and examined their transcriptional activity on the ARID5A promoter region. We found that all of the nonphospho-mimetic mutants, except for T749A, retained transcriptional activity on ARID5A promoter comparable to that of WT STAT1 (Fig. 3I). Moreover, converting the alanine residue in the T749A STAT1 mutant to a glutamic acid residue to generate the phospho-mimetic mutant T749E recapitulated the transcriptional activity of STAT1 on the ARID5A promoter (Fig. 3J). We next performed computational analysis to assess the accessibility of Thr749 within the STAT1 protein. We used a short Protein Data Bank (PDB) structure (2KA6 chain B), which spans the C terminus of STAT1 from amino acid residues 710 to 750 as one of the templates to construct a complete STAT1 homology model, based on which we found that Thr749 is possibly accessible similar to the canonical Tyr701 and Ser727 sites (Fig. 3K). Moreover, previous high-throughput, in-depth human proteomic and phospho-proteomics analyses annotated Thr749 as a phosphorylation site of STAT1 (33). Together, our data suggest that the noncanonical phosphorylation of Thr749 of STAT1 promotes ARID5A transcription.

Phosphorylation of STAT1 at Thr749 facilitates its binding to a noncanonical DNA motif in the ARID5A promoter region

Our observations raised the questions of whether Thr749 phosphorylation promotes STAT1 nuclear translocation, DNA binding, or both. To address these questions, we generated U3A cells lacking STAT1 to genetically overexpress different enhanced green fluorescent protein (EGFP)–tagged STAT1 constructs (referred hereafter as U3As1 cells) using a retrovirus-based transduction approach. We next used these U3As1 cells to monitor the nuclear translocation of different EGFP-tagged STAT1 constructs. We found that the tagged T749A and T749E STAT1 mutants translocated to the nucleus similarly to WT STAT1 (Fig. 4, A and B), suggesting that Thr749 is unlikely to regulate STAT1 nuclear translocation.

Fig. 4 Phosphorylation of STAT1 at Thr749 facilitates its binding to a noncanonical DNA motif in the ARID5A promoter region.

(A and B) U3As1 cells expressing EGFP-tagged STAT1-WT, STAT1-T749A, or STAT1-T749E. (A) Representative fluorescence micrographs showing the intracellular distribution of the indicated EGFP-tagged STAT1 proteins and the localization of the corresponding Hoechst-stained nuclei. Scale bar, 10 μm. Data are representative of three independent experiments. (B) Cellular cytoplasmic and nuclear fractions were isolated and analyzed by Western blotting to detect the indicated proteins. Blots are representative of three independent experiments. (C) Scheme of the full-length human ARID5A promoter (FULL) and its deletion construct (DEL). (D) Measurement of the luciferase activity of U3A cells 48 hours after transfection with luciferase reporter plasmids containing the promoter regions described in (C) together with control plasmid (EV) or an expression plasmid for WT STAT1. Results are presented relative to Renilla luciferase activity. Data are representative of three independent experiments and are presented as means ± SD. (E) Binding assay of STAT1 to biotinylated nucleotides containing a noncanonical DNA motif (5′-TTTGAGGC-′3) of the human ARID5A promoter. Cleared lysates from U3A cells expressing STAT1-WT, STAT1-T749A, or STAT1-T749E were incubated with the indicated biotinylated nucleotides. Bound proteins were immunoprecipitated with streptavidin beads and analyzed by Western blotting to detect STAT1. Input samples were prepared before incubation with biotinylated nucleotides. Blots are representative of three independent experiments. (F and G) U3A cells or U3As1 cells expressing STAT1-WT, STAT1-DoMut, STAT1-T749A, or STAT1-T749E. (F) Whole-cell lysates were separated by SDS-PAGE, and the indicated proteins were analyzed by Western blotting. Blots are representative of three independent experiments. (G) Total RNA was isolated, and the indicated transcripts were quantified by qRT-PCR analysis. Data are representative of three independent experiments and are presented as means ± SD. (H and I) U3As1 cells expressing STAT1-WT, STAT1-DoMut, STAT1-T749A, or STAT1-T749E. (H) Cells were stimulated with human IFN-γ (10 ng/ml) for 1 hour. Whole-cell lysates were separated by SDS-PAGE, and the indicated proteins were analyzed by Western blotting. Blots are representative of three independent experiments. (I) ChIP was performed with anti-STAT1 or control immunoglobulin G (IgG), which was followed by qPCR analysis with primers specific for the indicated promoter region of ARID5A. Fold enrichment was calculated using the percentage input method, and the graph shows the relative fold enrichment from the qPCR results. Data are representative of three independent experiments and are presented as means ± SD. (J and K) WT dTHP-1 cells were left unstimulated or were stimulated with LPS (100 ng/ml) for 4 hours. (J) Whole-cell lysates were harvested and cleared, and the pTyr- or pThr-containing proteins were subjected to immunoprecipitation. Immunoprecipitates from the stimulated samples were left untreated or were treated with lambda protein phosphatase (λPP) and analyzed by Western blotting. Input samples were prepared before the pTyr- or pThr-containing proteins were immunoprecipitated. Blots are representative of three independent experiments. (K) ChIP was performed using anti-STAT1 or control IgG, which was followed by qPCR analysis with primers specific for the indicated promoter regions of ARID5A or IRF1. Fold enrichment was calculated using the percentage input method, and the graph shows the relative fold enrichment from the qPCR assay results. Data are representative of three independent experiments and are presented as means ± SD. *P < 0.05, **P < 0.01, ***P < 0.001 as measured by one-way ANOVA with post hoc Tukey’s test (D, G, and I) or by paired Student’s t test (K).

We next considered the role of pThr749 on the ability of STAT1 to bind to DNA. We assessed the transcriptional activities of the NoTAD, WT, DoMut, and T749A STAT1 proteins on IRF1 and MX2 promoters, which contain a GAS and an ISRE, respectively (9). We found that the T749A STAT1 mutant, unlike the NoTAD and DoMut proteins, retained its transcriptional activity on the IRF1 and MX2 promoters (fig. S4A). We next found that Thr749 phosphorylation was dispensable for the formation of the ISGF3 complex (fig. S4B). Functionally, an ISGF3 complex containing the T749A STAT1 mutant showed similar transcriptional activity on the MX2 promoter to an ISGF3 complex containing WT STAT1 (fig. S4C). In contrast, STAT2 and IRF9 alone or together with the T749A mutant did not exert transcriptional activity on the ARID5A promoter (fig. S4C). These findings suggest that the phosphorylation of Thr749 of STAT1 confers STAT1 with a distinct DNA binding specificity.

To further investigate potential STAT1-binding sites in the ARID5A promoter, we used the JASPAR transcription factor binding site prediction tool (34). We next performed deletion analysis of the ARID5A promoter to identify which DNA region was critical for STAT1 binding (fig. S4D). Note that STAT1 did not show transcriptional activity in the JASPAR-predicted regions of the ARID5A promoter, whereas it showed transcriptional activity in an unpredicted region (−743 to −404) (fig. S4E). To identify the precise DNA binding motif for STAT1 in the ARID5A promoter, we performed another series of deletion analysis of the (−743 to −404) region of the ARID5A promoter (fig. 4F). We found that the 5′-TTTGAGGC-3′ nucleotide sequence within the (−743 to −404) region was critical for STAT1 transcriptional activity (fig. S4G). Consistently, deletion of 5′-TTTGAGGC-3′ from the full-length ARID5A promoter abolished STAT1 transcriptional activity (Fig. 4, C and D), indicating that STAT1 bound to this noncanonical sequence. We next performed a pull-down assay with bio-oligos containing the sequence 5′-TTTGAGGC-3′ to examine the physical interaction of STAT1 with this previously uncharacterized nucleotide sequence and found that STAT1 could bind (fig. S4H). In contrast, the T749A STAT1 mutant could not bind to bio-oligos containing the 5′-TTTGAGGC-3′ sequence, whereas it retained its binding to bio-oligos containing the GAS motif (Fig. 4E and fig. S4I). On the other hand, the T749E STAT1 mutant bound to bio-oligos containing the 5′-TTTGAGGC-3′ sequence (Fig. 4E).

We next examined whether the T749A mutation in STAT1 affected the expression of ARID5A and IL6 in U3As1 cells. We found that U3As1 cells expressing WT, DoMut, or the T749E STAT1 mutant had increased amounts of ARID5A and IL6 mRNAs compared to U3A cells or U3As1 cells expressing the T749A STAT1 mutant (Fig. 4, F and G). To confirm that ARID5A expression required STAT1-pThr749, but not STAT1-pTyr701, we analyzed ARID5A expression in U3As1 cells upon stimulation with recombinant human IFN-γ. We found that IFN-γ–stimulated U3As1 expressing the DoMut STAT1 mutant did not generate STAT1-pTyr701, whereas they had a similar amount of ARID5A protein to that in U3As1 cells expressing WT STAT1. In contrast, IFN-γ–stimulated U3As1 expressing the T749A STAT1 mutant did not express ARID5A protein, whereas they had a similar amount of STAT1-pTyr701 to that in U3As1 cells expressing WT STAT1 (Fig. 4H). We next performed chromatin immunoprecipitation–quantitative polymerase chain reaction (ChIP-qPCR) analysis to assess the binding of different STAT1 mutants to the ARID5A promoter in U3As1 cells. Consistent with our earlier findings, statistically significantly less T749A STAT1 mutant was bound to the ARID5A promoter region containing the 5′-TTTGAGGC-3′ sequence compared to the amounts of WT, DoMut, and T749E STAT1 mutant that bound to the same sequence (Fig. 4I).

We next examined whether endogenous STAT1 was threonine phosphorylated and activated the ARID5A promoter in LPS-stimulated human macrophages. We found that LPS stimulation of human macrophages induced the threonine phosphorylation of STAT1 and that the treatment of immunoprecipitated STAT1 with lambda protein phosphatase (λPP) abolished this phosphorylation (fig. S4J). Moreover, we observed STAT1 in samples immunoprecipitated with anti–phospho-tyrosine (anti-pTyr) or anti–phospho-threonine (anti-pThr) antibodies from LPS-stimulated cells, which was abolished by λPP (Fig. 4J). We next performed ChIP-qPCR analysis to assess whether STAT1 bound to the ARID5A promoter in LPS-stimulated macrophages. We found that, upon LPS stimulation, STAT1 bound to both the canonical GAS motif of the IRF1 promoter region and the ARID5A promoter region containing the 5′-TTTGAGGC-3′ sequence (Fig. 4K). Together, our results suggest that pThr749 facilitates the binding of STAT1 to a noncanonical DNA motif (5′-TTTGAGGC-3′), resulting in the transcriptional activation of ARID5A.

A noncanonical TBK1-IKKβ heterodimer mediates the phosphorylation of STAT1 at Thr749 downstream of TLR4 endocytosis

We next sought to identify the kinase that mediated the phosphorylation of STAT1 at Thr749. TLR4 endocytosis activates TBK1 and IKKɛ, which are well-known serine and threonine kinases downstream of TRIF (35). Thus, we examined whether TBK1 or IKKɛ mediated the phosphorylation of STAT1 Thr749 and subsequently promoted ARID5A expression. We found that the TBK1/IKKɛ chemical inhibitor BX795 statistically significantly reduced ARID5A expression in LPS-stimulated MDMs (Fig. 5A, dark gray bars). Our group previously showed that IKKβ is critical for Arid5a expression by LPS-stimulated mouse embryonic fibroblasts, whereas canonical, IKKβ-mediated phosphorylation of p65 at Ser536 was dispensable (36). We therefore examined whether IKKβ promoted the phosphorylation of STAT1-Thr749 and the subsequent expression of ARID5A. We found that the IKKβ chemical inhibitor BMS-345541 statistically significantly inhibited ARID5A expression in LPS-stimulated MDMs (Fig. 5A, light gray bars). We next performed siRNA-mediated knockdown of IKKα, IKKβ, IKKε, or TBK1 and assessed ARID5A expression in LPS-stimulated dTHP-1 cells. We observed that the knockdown of either IKKβ or TBK1 inhibited ARID5A expression, whereas knockdown of IKKα or IKKε had no effect (fig. S5, A to E). We hypothesized that IKKβ and TBK1 might form a heterodimer that mediated the phosphorylation of STAT1-Thr749. Consistent with previous studies (2), we found that at the early phase (0.5 to 2 hours) of LPS stimulation, IKKβ formed a complex with IKKα. Notably, in the late phase (3 to 6 hours) of LPS stimulation, IKKβ formed a heterodimer with TBK1 (Fig. 5B and fig. S5F). Note that the IKKβ-TBK1 complex interacted with STAT1 and that treatment of the immunoprecipitated STAT1 with λPP abolished its interaction with IKKβ and TBK1 (Fig. 5C), suggesting that the IKKβ-TBK1 complex phosphorylates STAT1 upon LPS stimulation. We next examined whether the IKKβ-TBK1–mediated phosphorylation of STAT1 depended on TLR4 endocytosis in human macrophages. We found that the interaction between STAT1 and the IKKβ-TBK1 complex, as well as its threonine phosphorylation, was unabated in MYD88−/− and IFNAR2−/− dTHP-1 cells, whereas treatment with MitMAB diminished both the interaction and the phosphorylation event (Fig. 5D). These findings suggest that LPS-induced TLR4 endocytosis promotes the threonine phosphorylation of STAT1 by the IKKβ-TBK1 complex.

Fig. 5 A noncanonical TBK1-IKKβ heterodimer promotes the phosphorylation of STAT1 at Thr749 downstream of TLR4 endocytosis.

(A) MDMs were stimulated with LPS (100 ng/ml) in the presence or absence of 10 μM BX795 or 10 μM BMS-345541. Total RNA was isolated 3 hours after stimulation, and ARID5A transcripts were quantified by qRT-PCR analysis. Data are representative of three independent experiments (five donors per experiment) and are presented as means ± SD. (B and C) WT dTHP-1 cells were left unstimulated or were stimulated with LPS (100 ng/ml). (B) Whole-cell lysates were harvested at the indicated times. IKKβ was immunoprecipitated from cleared lysates. IKKβ immunoprecipitates were subjected to Western blotting analysis of the indicated endogenous proteins. Input samples were prepared before IKKβ immunoprecipitation was performed. (C) Whole-cell lysates were harvested 4 hours after stimulation, and STAT1 was then immunoprecipitated. STAT1 immunoprecipitates from stimulated samples were left untreated or were treated with λPP and then analyzed by Western blotting for the indicated endogenous proteins. Input samples were prepared before STAT1 immunoprecipitation was performed. Blots are representative of three independent experiments. (D) WT, MYD88−/−, or IFNAR2−/− dTHP-1 cells were stimulated with LPS (100 ng/ml) for 4 hours in the presence or absence of 10 μM MitMAB. Whole-cell lysates were harvested, and IKKβ, TBK1, and pThr-containing proteins were immunoprecipitated from the cleared lysates. Immunoprecipitated samples were analyzed by Western blotting to detect the indicated endogenous proteins. Input samples were prepared before the immunoprecipitations were performed. Blots are representative of three independent experiments. (E and F) U3A cells were transiently co-transfected with the indicated expression plasmids. Twenty-four hours later, HA-tagged proteins were immunoprecipitated from the cleared cell lysates and then analyzed by Western blotting to detect the indicated proteins. Input samples were prepared before immunoprecipitations were performed. Blots are representative of three independent experiments. (G) U3A cells were transiently co-transfected with the indicated expression plasmids. Twenty-four hours later, Flag-tagged proteins were immunoprecipitated from the cleared cell lysates and then analyzed by Western blotting to detect the indicated proteins. Input samples were prepared before immunoprecipitations were performed. Blots are representative of three independent experiments. (H and I) U3As1 cells expressing STAT1-WT or STAT1-T749A were transiently transfected with plasmid expressing IKKβ and then cultured for 24 hours. (H) Total RNA was isolated, and the indicated transcripts were quantified by qRT-PCR analysis. Data are representative of three independent experiments and are presented as means ± SD. (I) Whole-cell lysates were separated by SDS-PAGE, and the indicated proteins were analyzed by Western blotting. Blots are representative of three independent experiments. (J) Scheme of the full-length STAT1 (FULL) and STAT1 (G3S) constructs. (K and L) U3A cells were transiently co-transfected with the indicated expression plasmids. Twenty-four hours later, HA-tagged (K) or Flag-tagged (L) proteins were immunoprecipitated from the cleared cell lysates and analyzed by Western blotting to detect the indicated proteins. Input samples were prepared before the immunoprecipitations were performed. Blots are representative of three independent experiments. **P < 0.01, ***P < 0.001, ****P < 0.0001 as measured by one-way ANOVA with post hoc Tukey’s test (A and H).

We next investigated which kinase in the IKKβ-TBK1 complex mediated the threonine phosphorylation of STAT1 and whether this phosphorylation occurred at Thr749. Tojima et al. (37) demonstrated that TBK1 activates IKKβ in vitro through the phosphorylation of the residues Ser177 and Ser181. We therefore examined whether TBK1 activated IKKβ, which, in turn, phosphorylated STAT1 at Thr749. We generated kinase-defective mutants of IKKβ and TBK1 (the K44A and K38A variants, respectively) and examined whether either the IKKβ K44A mutant or the TBK1 K38A mutant impaired the interaction between the IKKβ-TBK1 heterodimer and STAT1. We found that the IKKβ K44A mutant did not bind to STAT1, whereas it retained its ability to bind to TBK1 (Fig. 5E). In contrast, the TBK1 K38A mutant did not affect the interaction between IKKβ and STAT1, whereas it abolished the interaction between TBK1 and IKKβ (Fig. 5E). Moreover, TBK1 did not bind very efficiently to the IKKβ S177A/S181A mutant (fig. S5G).

We next examined whether IKKβ directly interacted with and phosphorylated Thr749 of STAT1. We found that IKKβ interacted with WT and DoMut STAT1, whereas it did not interact with the T749A STAT1 mutant (Fig. 5F). We next tested whether IKKβ phosphorylated STAT1 at Thr749. We found that overexpression of IKKβ induced the threonine phosphorylation of STAT1 immunoprecipitated from cells expressing WT STAT1 but not the T749A mutant (Fig. 5G and fig. S5H). In addition, overexpressing IKKβ in U3As1 cells resulted in the slower migration of WT STAT1, but not the T749A STAT1 mutant, in a Phos-tag acrylamide gel (fig. S5I). Furthermore, we found that the IKKβ K44A kinase-defective mutant did not promote the threonine phosphorylation of STAT (Fig. 5G). Functionally, overexpressing IKKβ resulted in the enhanced expression of ARID5A and IL6 in U3As1 expressing WT STAT1, but not in those expressing the T749A STAT1 mutant (Fig. 5, H and I). In addition, we generated a STAT1 construct lacking the TAD and linked to amino acid residues 742 to 750 through five repeats of flexible glycine-serine (G3S) (referred hereafter as STAT1G3S) (Fig. 5J). In experiments with STAT1G3S, we excluded any possible role of the canonical Tyr701 and Ser727 phosphorylation sites and enhanced the sensitivity and the specificity of our assay. We found that IKKβ interacted with both full-length STAT1 and STAT1G3S (Fig. 5K). Moreover, the K44A mutation of IKKβ or the T749A mutation of STAT1G3S (STAT1G3S/T749A) reduced the interaction between IKKβ and STAT1 G3S (Fig. 5L). Together, our data suggest that the IKKβ kinase domain conducts the phosphorylation of STAT1 at Thr749.

LPS-induced TLR4 endocytosis augments IL12B transcription in a manner dependent on STAT1-pThr749

We next investigated whether STAT1-pThr749 augmented expression of the gene encoding IL-12p40 downstream of TLR4 endocytosis. We first examined different proinflammatory cytokines in STAT1 knockdown dTHP-1 cells upon LPS stimulation. We found that STAT1 knockdown led to the statistically significantly decreased expression of IL6 and IL12B, but not TNF (Fig. 6A). Moreover, MitMAB treatment attenuated IL6 and IL12B expression upon LPS stimulation of shScramble dTHP-1 cells but not STAT1 knockdown dTHP-1 cells (Fig. 6B), indicating that TLR4 endocytosis augments IL12B expression through STAT1.

Fig. 6 LPS-induced TLR4 endocytosis augments IL12B transcription in a manner dependent on STAT1-pThr749.

(A) Transduced dTHP-1 cells expressing the indicated shRNAs were left unstimulated or were stimulated with LPS (100 ng/ml) for the indicated times. Total RNA was isolated, and the indicated transcripts were quantified by qRT-PCR analysis. Data are representative of three independent experiments and are presented as means ± SD. (B) Transduced dTHP-1 cells expressing the indicated shRNAs were stimulated with LPS (100 ng/ml) in the presence or absence of 10 μM MitMAB. Total RNA was isolated 3 hours after stimulation, and the indicated transcripts were quantified by qRT-PCR analysis. Data are representative of three independent experiments and are presented as means ± SD. (C) Scheme of the full-length human IL12B promoter (FULL) and its deletion construct (DEL). (D) Measurement of the luciferase activity of U3A cells 48 hours after transfection with luciferase reporter plasmids containing the promoter regions described in (C) together with control plasmid (EV) or expression plasmid for WT STAT1. Results are presented relative to Renilla luciferase activity. Data are representative of three independent experiments and are presented as means ± SD. (E) Measurement of the luciferase activity of U3A cells 48 hours after transfection with a luciferase reporter plasmid containing the full-length human IL12B promoter, together with control plasmid (EV) or expression plasmid for WT or T749A STAT1. Results are presented relative to Renilla luciferase activity. Data are representative of three independent experiments and are presented as means ± SD. (F) U3A or U3As1 cells expressing STAT1-WT, STAT1-DoMut, STAT1-T749A, or STAT1-T749E. Total RNA was isolated, and IL12B transcripts were quantified by qRT-PCR analysis. Data are representative of three independent experiments and are presented as means ± SD. (G) U3As1 cells expressing STAT1-WT, STAT1-DoMut, STAT1-T749A, or STAT1-T749E. ChIP was performed using anti-STAT1 or control IgG, which was followed by qPCR analysis with primers specific for the indicated IL12B promoter region. Fold enrichment was calculated using the percentage input method, and the graph shows the relative fold enrichment from the qPCR assay results. Data are representative of three independent experiments and are presented as means ± SD. (H) U3As1 cells expressing STAT1-WT or STAT1-T749A were transiently transfected with expression plasmid for HA-IKKβ. Twenty-four hours later, total RNA was isolated and IL12B transcripts were quantified by qRT-PCR analysis. Data are representative of three independent experiments and are presented as means ± SD. (I) WT dTHP-1 cells were left unstimulated or were stimulated with LPS (100 ng/ml) for 4 hours. ChIP was performed with anti-STAT1 or control IgG, which was followed by qPCR analysis with primers specific for the indicated promoter regions of IL12B. Fold enrichment was calculated using the percentage input method, and the graph shows the relative fold enrichment from the qPCR assay results. Data are representative of three independent experiments and are presented as means ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 as measured by one-way ANOVA with post hoc Tukey’s test (A, B, and D to H) or by paired Student’s t test (I).

We next examined whether STAT1 directly bound to and activated the IL12B promoter. We observed that the IL12B promoter contains a noncanonical motif (5′-TTTGAGTC-3′) similar to that of ARID5A (Fig. 6C). We found that STAT1 exerted transcriptional activity on the IL12B promoter, and that deletion of the 5′-TTTGAGTC-3′ sequence from the promoter region abolished STAT1 activity (Fig. 6D). We next examined whether STAT1-pThr749 facilitated the transcriptional activity of STAT1 on the IL12B promoter. We found that the T749A STAT1 mutant did not promote the transcriptional activity on the IL12B promoter (Fig. 6E). In addition, we found that U3As1 cells expressing WT STAT1, DoMut STAT1, or the T749E STAT1 mutant expressed greater amounts of IL12B transcripts compared to those in U3A or U3As1 cells expressing the T749A STAT1 mutant (Fig. 6F). We next performed ChIP-qPCR analysis to assess the binding of different STAT1 mutants to the IL12B promoter in U3As1 cells. The T749A mutant showed attenuated STAT1 binding to the IL12B promoter region containing the 5′-TTTGAGTC-3′ sequence compared to WT STAT1, DoMut STAT1, or the T749E STAT1 mutant (Fig. 6G). Moreover, overexpressing IKKβ enhanced IL12B expression in U3As1 cells expressing WT STAT1, but not in those expressing the T749A STAT1 mutant (Fig. 6H). We next examined whether endogenous STAT1 bound to the IL12B promoter in LPS-stimulated human macrophages. We found that upon LPS stimulation, STAT1 bound to the IL12B promoter region containing the 5′-TTTGAGTC-3′ sequence as measured by ChIP-qPCR (Fig. 6I). Together, our data indicate that TLR4 endocytosis enhances the transcription of IL12B through STAT1-pThr749.

DISCUSSION

Proinflammatory cytokines are pivotal not only for eliminating bacterial infections but also for driving unwanted inflammation in lethal sepsis. Therefore, coordinated circuits of transcriptional and posttranscriptional regulators exist to prevent their aberrant production. Our group and others have shown that Arid5a stabilizes IL6 mRNA (23, 38). Arid5a counteracts the destabilizing effect of Regnase-1 on IL6 mRNA, and tight regulation of the balance between Regnase-1 and Arid5a is critical for determining the half-life of IL6 mRNA and preventing the aberrant production of IL-6 (39). Here, we found that TLR4 endocytosis promoted ARID5A production, which subsequently stabilized IL6 mRNA in human macrophages. Three independent lines of evidence supported our conclusion. First, chemical inhibition of TLR4 endocytosis reduced ARID5A and IL-6 production in response to LPS. Second, stimulation of TLR2, which lacks an endosomal TRIF component, failed to induce ARID5A expression and resulted in reduced IL-6 production compared to that induced by TLR4 stimulation. Third, MYD88−/− macrophages retained their ability to express ARID5A transcripts and protein, whereas TRIF knockdown abolished ARID5A expression. Our identification of ARID5A as a proinflammatory effector downstream of TLR4 endocytosis provides a likely explanation of previous reports that showed increased IL-6 production upon TLR4 stimulation as compared to TLR2 stimulation in Regnase1-deficient murine macrophages, as well as the enhanced sepsis-induced mortality observed in response to TLR4 stimulation compared to that in response to TLR2 stimulation (40, 41).

Although TLR4 endocytosis is a key step for IFN-β production, the IFN-β–JAK–STAT1-pTyr701 signaling pathway is unlikely to contribute to proinflammatory cytokine production. Many reports have shown that deficiencies in IFN-β, IFNAR, or Tyk2 do not affect the production of proinflammatory cytokines, such as IL-6, in response to bacterial infections (1113). In contrast, STAT1 deficiency results in diminished production of IL-6 and enhanced survival in response to bacterial infections (42, 43). These observations suggest that the role of STAT1 extends beyond that of JAK–STAT1-pTyr701 signaling. This perspective has been highlighted by reports showing that STAT1 exerts distinct functions independently of STAT1-pTyr701 signaling (32, 44). Supporting this premise, our work demonstrates that STAT1 rather than STAT1-pTyr701 promoted ARID5A expression. Three pieces of evidence supported our conclusion. First, LPS-stimulated IFNAR2−/− dTHP-1 cells did not activate STAT1-pTyr701 at the time when they expressed ARID5A transcripts and protein at amounts comparable to those in WT cells. By contrast, knocking down STAT1 in IFNAR2 −/− dTHP-1 cells reduced the amounts of ARID5A transcripts and protein. Second, human IFN-β–stimulated MDMs exhibited the phosphorylation of STAT1 at pTyr701, but did not induce the expression of ARID5A. Third, the siRNA-mediated knockdown of IRF3 abolished the production of IFN-β and the phosphorylation of STAT1 at Tyr701, but did not affect the expression of ARID5A in human macrophages. The simplest interpretation of these observations is that STAT1 function extends beyond its role in mediating canonical JAK–STAT1-pTyr701 signaling.

Consistent with previous work (43), we found that STAT1 deficiency diminished the expression of antiviral response genes, whereas NF-κB activation and TNF-α production were unabated. On the other hand, we observed that STAT1 knockdown substantially reduced IL-6 and IL-12p40 production. These observations, coupled with previous reports showing that TNF-α production does not require TLR4 endocytosis (4, 5), suggest a possible proinflammatory role of STAT1 downstream of TLR4 endocytosis. Consistent with this idea, we found that MitMAB did not decrease the production of IL-6 and IL-12p40 in STAT1 knockdown macrophages. Our work revealed that TLR4 endocytosis promoted noncanonical STAT1-pThr749 signaling, which promoted ARID5A and IL12B transcription. Note that the phosphorylation of Thr749 did not affect the nuclear translocation of STAT1. Instead, it facilitated STAT1 binding to and activation of the ARID5A and IL12B promoters. Four independent experimental approaches supported our conclusion. First, whereas the T749A STAT1 mutant retained its transcriptional activity on IRF1 and MX2 promoters, it failed to induce ARID5A promoter activity as measured by luciferase assay. Moreover, the T749E mutant rescued STAT1 transcriptional activity on the ARID5A promoter. Second, WT STAT1 and the T749E STAT1 mutant bound to oligonucleotides containing GAS or the noncanonical 5′-TTTGAGGC-3′ sequence, whereas the T749A STAT1 mutant could only bind to the GAS sequence, but not the 5′-TTTGAGGC-3′ sequence. Third, U3As1 cells expressing WT STAT1 or the T749E STAT1 mutant expressed statistically significantly greater amounts of ARID5A and IL12B transcripts compared to those of cells expressing the T749A STAT1 mutant. Fourth, WT STAT1 and the T749E STAT1 mutant showed statistically significantly greater binding to the ARID5A and IL12B promoters compared to that of the T749A STAT1 mutant in U3As1 cells. In addition, in LPS-stimulated macrophages, endogenous STAT1 could bind to the ARID5A and IL12B promoters containing the 5′-TTTGAGGC-3′ or 5′-TTTGAGTC-3′ sequences, respectively. Together, our data suggest that the different phosphorylation of STAT1 confers distinct DNA binding and gene regulation properties, where canonical phosphorylation of Tyr701 and Ser727 facilitates STAT1-dependent expression of antiviral genes. On the other hand, the phosphorylation of Thr749 confers STAT1 with a proinflammatory function, which expands its role into regulating IL12B and IL6 transcription and mRNA stabilization, respectively.

Our mechanistic analysis of the different stages of TLR4 signaling highlights the fundamental role of the spatiotemporal regulation of TLR4 in regulating the expression of proinflammatory cytokines. At the early phase of LPS stimulation, signaling by TLR4-MYD88 at the cell surface through the IKKα-IKKβ complex promotes NF-κB activation and Regnase-1 degradation, which, in turn, promote IL-6 and IL-12p40 production (45). On the other hand, TLR4 endocytosis prolongs the expression of these cytokines in the late phase of LPS stimulation by stimulating the phosphorylation of STAT1 at Thr749 through the TBK1-IKKβ complex. We found that TBK1, but not IKKε, interacted with and activated IKKβ in the late phase of LPS stimulation. Subsequently, IKKβ, but not TBK1, mediated the phosphorylation of STAT1 at Thr749. These findings are consistent with the longer half-life of TBK1 compared to that of IKKε and the fact that TBK1 interacts with and activates IKKβ (37, 46). In addition, we also demonstrated that the canonical phosphorylation of STAT1 at Tyr701 and Ser727 was dispensable for the IKKβ-mediated phosphorylation of STAT1 at Thr749. Although our findings suggest that STAT1 serves as a proinflammatory mediator at the late phase of TLR4 signaling, it is not unlikely that other transcriptional factors interact with STAT1 for optimal regulation. For example, a cooperative interaction between STAT1 and NF-κB has been reported for the regulation of multiple proinflammatory mediators (4750). In this regard, our group previously showed that Arid5a expression does not require the canonical phosphorylation of NF-kB at Ser536. Rather, it requires the acetylation of NF-κB at Lys310, Lys314, and Lys315 (36), which occurs during the late phase of LPS stimulation (51).

In summary, our study unveils a proinflammatory branch of STAT1 signaling, which provides a potential mechanistic explanation for the longstanding question of how endosomal TLR4-TRIF signaling regulates proinflammatory cytokine expression independently of NF-κB activation. Future experiments will be required to better understand how the phosphorylation status of STAT1 determines its DNA transcriptional activity and whether other DNA binding proteins are required for such activity.

MATERIALS AND METHODS

Cell lines

Human pre-monocyte cells, THP-1 cells (THP-1 Dual, InvivoGen, thpd-nfis), MYD88−/− THP-1 cells (THP-1 Dual KO-Myd, Invivogen, thpd-komyd), and IFNAR2−/− THP-1 cells (THP-1 Dual KO-IFNAR2, InvivoGen, thpd-koifnar2) were cultured in RPMI 1640 medium (Nacalai Tesque) supplemented with 10% fetal calf serum (FCS; Sigma), streptomycin (100 μg/ml), and penicillin G (100 U/ml; Nacalai Tesque). Human embryonic kidney (HEK) 293T cells (American Type Culture Collection, CRL-3216), U3A cells [European Collection of Authenticated Cell Cultures (ECACC), 12021503], and U3As1 cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) (Nacalai Tesque) supplemented with 10% FCS (Sigma), streptomycin (100 μg/ml), and penicillin G (100 U/ml; Nacalai Tesque).

Differentiation of THP-1 cells into macrophage-like cells (dTHP-1 cells)

THP-1 cells cultured in RPMI 1640 supplemented with 10% FCS, streptomycin (100 μg/ml), and penicillin G (100 U/ml) were treated with phorbol 12-myristate 13-acetate (PMA) (InvivoGen, tlrl-pma) at a final concentration of 200 nM for 48 hours. The cells were then washed with phosphate-buffered saline (PBS) once and cultured in fresh RPMI 1640 supplemented with 10% FCS, streptomycin (100 μg/ml), and penicillin G (100 U/ml) for 72 hours.

Preparation of human MDMs

Blood samples (30 to 50 ml) from healthy donors were centrifuged at 260g at 4°C for 5 min to separate blood cells from plasma. The plasma layer was collected and heat-inactivated at 56°C for 30 min. Inactivated plasma was centrifuged at 540g at 4°C for 5 min, and then the supernatant was filtrated through a 0.45-μm strainer. Separated blood samples were diluted with an equal volume of RPMI 1640 and separated by gradient (Histopaque-1077, Sigma) through centrifugation at 260g at 25°C for 45 min, and then a buffy coat containing PBMCs was isolated. The number of PBMCs was counted, and monocytes were isolated with CD14-positive magnetic beads (human CD14 microbeads, Miltenyi Biotec, 130-050-201) according to the manufacturer’s protocol. In brief, 1 × 107 PBMCs were incubated with 20 μl of CD14-positive magnetic beads on ice for 15 min and then were isolated with an autoMACS Pro Separator (Miltenyi Biotec) by positive selection. Isolated human monocytes were differentiated into MDMs in RPMI 1640 containing 10% plasma from the same donor and recombinant human granulocyte-macrophage colony-stimulating factor (GM-CSF) (PeproTech, 300-03) at final concentration of 5 ng/ml for 7 days.

LPS, Pam3CSK4, and IFN treatments

Confluent cell monolayers were stimulated with the concentrations of LPS (Escherichia coli O55:B5, Sigma-Aldrich, L4524), Pam3CSK4 (InvivoGen, tlrl-pms), or recombinant human IFN-β (R&D Systems, 8499-IF) indicated in the figure legends.

Chemical inhibitors

Macrophages were pretreated with MitMAB (Abcam, ab120466) or dynasore (Abcam, 120192) (at the concentrations indicated in the figure legends) in serum-free RPMI 1640 containing recombinant human LPS-binding protein (R&D Systems, 870-LP-025) at a final concentration of 50 ng/ml for 30 min before and then throughout the indicated stimulations. Macrophages were pretreated with the concentrations of tofacitinib citrate (Sigma-Aldrich, PZ0017) indicated in the figure legends in RPMI 1640 supplemented with 10% FCS, streptomycin (100 μg/ml), and penicillin G (100 U/ml) for 30 min before and throughout the indicated stimulations.

siRNA-mediated knockdowns

Cells were plated at 70 to 80% confluency per well of a 24-well dish. Cells were transfected with 100 nM of the appropriate siRNA with the Lipofectamine RNAiMAX Transfection Reagent (Thermo Fisher Scientific, 13778150) in case of THP-1 cells or with Viromer Green (Lipocalyx, VG-01LB-01) in case of MDMs according to the manufacturer’s instructions. Depletion efficiency was determined 48 hours after transfection. All siRNAs used were purchased from Applied Biosystems as follows: ARID5A (#ARID5AHSS173931), TRIF (#TICAM1HSS152364), STAT1 (#STAT1VHS40871), IRF3 (#IRF3HSS105507), CHUK (#CHUKVHS40781), IKBKB (#IKBKBVHS40301), IKBKE (#IKBKEHSS114410), TBK1 (#TBK1HSS120901), Stealth RNAi siRNA Negative Control, Med GC (#12935300).

Generation of STAT1 knockdown THP-1 cells

HEK293T cells were cotransfected with the pLKO.1 puro plasmid encoding scrambled or different STAT1-specific shRNAs (Sigma-Aldrich; STAT1#1 TRCN0000004267 and STAT1#2 TRCN0000280024) with the appropriate packaging and envelop plasmids for 16 hours. The cell culture medium was changed, and viral particles were collected 48 hours later. THP-1 cells were incubated with the appropriate viral particles with polybrene (8 μg/ml; Nacalai Tesque) for 24 hours. Cells were centrifuged, and fresh medium was added including puromycin (1 μg/ml; Wako). Knockdown efficiency was confirmed by Western blotting analysis.

Generation of U3As1 cells

Retroviral particles containing pCX4-bsr-EGFP-Flag expressing the appropriate STAT1 constructs were prepared using the Retrovirus Packaging Kit Ampho (TakaraBio, 6161) according to the manufacturer’s protocol. U3A cells were incubated with the appropriate viral particles with polybrene (10 μg/ml; Nacalai Tesque) for 24 hours. Cells were washed, and fresh medium was added including blasticidin S (5 μg/ml; Wako). Overexpression efficiency was confirmed by detecting EGFP by Western blotting and immunofluorescence analyses.

RNA isolation and qRT-PCR analysis

Total RNA was isolated from cells with the RNeasy Mini Kit (Qiagen, 74106). Complementary DNA (cDNA) was synthesized with PrimeScript Reverse Transcriptase (TakaraBio, RR047A) according to the manufacturer’s protocol. All quantitative real-time PCRs (qRT-PCRs) were performed with a ViiA 7 Real-Time PCR System and Quanti Studio 3 (Applied Biosystems) using PowerUp SYBR Green Master Mix (Applied Biosystems, A25742) and custom-designed primers in a final volume of 10 μl. Cycling conditions were 98°C for 2 min, followed by 40 cycles of 98°C for 10 s and 60°C for 30 s. Relative expression of target genes was measured using the comparative ΔΔCt method, and human ACTB was used as internal control. The sequences of the qRT-PCR primers are listed in table S1.

Western blotting and immunoprecipitation

Whole-cell lysates were prepared using the RIPA Lysis Buffer System (Santa Cruz Biotechnology, sc-24948A). Cellular cytoplasmic and nuclear fractions were isolated with the Nuclear Extraction Kit (Abcam, ab113473). Protein concentration was quantified by DC protein assay (Bio-Rad, 500-0116). Lysates were boiled in SDS sample buffer (Nacalai Tesque) for 5 min at 96°C and resolved by 5 to 20% SDS-PAGE gels (Nacalai Tesque). To detect STAT1 phosphorylation status, SuperSep Phos-tag gels (WAKO, 195-17371) were used according to the manufacturer’s recommendations. Proteins were transferred onto 0.45-μm polyvinylidene difluoride membrane (GE Healthcare) by semi-dry transfer system using the Transfer-Blot SD (Bio-Rad). After transfer, the membrane was blocked in 5% (w/v) skim milk suspended in 1× tris-buffered saline (TBS) containing 0.1% Tween 20 (TBST) for 2 hours at room temperature. Membranes were washed with TBST and incubated with the appropriate primary antibodies in Can Get Signal Immunereaction Enhancer (TOYOBO, NKB-101 T) Solution 1 (diluted by 1:1000) overnight at 4°C. After primary antibody incubation, the membrane was washed three times with TBST and incubated with Amersham ECL horseradish peroxidase–conjugated anti-rabbit or anti-mouse immunoglobulin G (IgG) (GE Healthcare) in Can Get Signal Immunereaction Enhancer Solution 2 (diluted by 1:10,000) for 1 hour at room temperature. The membrane was then washed three times with TBST and developed with Chemi-Lumi One Ultra (Nacalai Tesque, 11644-40). Images were acquired with an Image Quant LAS500 (GE Healthcare). For immunoprecipitations, whole-cell lysates were precleared, and proteins were immunoprecipitated from the lysates overnight at 4°C with the appropriate antibodies or control IgG. Subsequently, Pierce Protein G Magnetic Beads (Thermo Fisher Scientific, 88848) were added for 2 hours at 4°C to collect immune complexes. Then, the protein G beads were washed four times with lysis buffer and suspended in SDS sample buffer. Immunoprecipitated samples were incubated for 5 min at 96°C and then subjected to Western blotting analysis. The antibodies used for Western blotting and immunoprecipitations are listed in table S2.

Flow cytometry analysis

LPS-stimulated dTHP-1 cells were harvested in 0.25% trypsin-EDTA (Nacalai Tesque), and then cell pellets were collected by centrifugation at 2400g at 4°C for 5 min. Cells were incubated with Human TruStain FcX (diluted by 1:20, BioLegend) in PBS with 2% FCS for 30 min on ice to block Fc receptors. Anti-human TLR4–phycoerythrin (PE) (1:100 dilution, BioLegend, clone HTA125, 312806) or PE-isotype control (1:100 dilution, BioLegend, 400212) was added to the cells and incubated for 30 min on ice. After washing twice with PBS, the cells were treated with propidium iodide (diluted by 1:400, Sigma-Aldrich) and subjected to flow cytometry analysis with a FACS CantoII flow cytometer (BD Pharmingen).

Enzyme-linked immunosorbent assay

Cytometric beads assay [Becton Dickinson (BD) Pharmingen] was used to measure the amounts of different human cytokines. Human Soluble Protein Master Buffer Kit (BD, 558264), Human IL-6 Flex Set (BD, 558276), Human IL-12/IL-23p40 Flex Set (BD, 650154), and the Human TNF-α Flex Set (BD, 560112) were used in flow cytometry analyses according to the manufacturer’s instructions.

Plasmid construction and mutagenesis

Recombinant overexpressing vectors encoding the tagged indicated human proteins were constructed with the In-Fusion HD Cloning kit (Clontech, 639648) according to the manufacturer’s protocol. Constructs expressing ARID5A (NM_001319085.1), IRF3 (NM_001197122.1), STAT1 (NM_007315), IKBKB (NM_001556), TBK1 (NM_013254), STAT2 (NM_005419.4), and IRF9 (ENST00000396864.7) were prepared by PCR-based amplification from human macrophage cDNA. The Flag-STAT1, hemagglutinin (HA)–IKKβ, and Myc-TBK1 mutant constructs were generated with the KOD-Plus-Mutagenesis Kit (TOYOBO, SNK-101) according to the manufacturer’s instructions. All constructs were confirmed by DNA sequencing and Western blotting analysis. Primers used for PCR-based amplification are listed in table S1.

Luciferase reporter assays

The appropriate promoter regions of the ARID5A, IRF1, MX2, and IL12B genes or the 3′UTR region of IL6 mRNA were amplified from human macrophage genomic DNA by PCR and subcloned into the pGL3 luciferase reporter vectors (Promega) using the In-Fusion HD kit according to the manufacturer’s protocol. Deletion constructs of the appropriate promoters were prepared with the KOD-Plus-Mutagenesis Kit according to the manufacturer’s instructions. U3A or HEK293T cells cultured in 24-well plates were cotransfected with the appropriate pGL3-luciferase plasmid or pGL3-luciferase control plasmid, together with the appropriate overexpression plasmid and the Renilla Luciferase control reporter (Promega). Transfections were performed with the Lipofectamine 3000 kit (Thermo Fisher Scientific, L3000015) or Lipofectamine LTX (Thermo Fisher Scientific, 15338100) for U3A and HEK293T cells, respectively. Forty-eight hours later, the luciferase activity in the cell lysates was measured with the Dual Luciferase Reporter Assay System, and the data were normalized to Renilla luciferase activity. Primers used for PCR-based amplification are listed in table S1.

Immunofluorescence

U3As1 cells expressing the appropriate EGFP-tagged STAT1 constructs were cultured in μ-Slide 8 Well (Ibidi). Cells were fixed in 4% paraformaldehyde (Nacalai Tesque) in PBS for 10 min at room temperature, and the nuclei were stained for 10 min with NucBlue Live Ready Probes Reagent (Thermo Fisher Scientific, R37605). Samples were mounted in SlowFade Diamond Antifade Mountant (Thermo Fisher Scientific, S36967) and visualized with the 100× lens of a BZ-X710 microscope (Keyence) equipped with the appropriate fluorescence filters.

Pull-down assays with biotinylated oligonucleotides

Biotin end-labeled oligonucleotides were conjugated to Dynabeads M-280 Streptavidin (Thermo Fisher Scientific, 11205D) for 30 min at room temperature according to the manufacturer’s protocol. Conjugated beads were incubated for 2 hours at 4°C with cleared lysates from confluent cells transfected with the appropriate overexpression plasmids. The beads were then washed four times with wash buffer, and the bound proteins were eluted by 95% formamide, 10 mM EDTA, and free biotin at 65°C for 10 min and then analyzed by Western blotting. The oligonucleotides used are as follows: hIL6 3′UTR, 5′-biotin-UAUUUUAAUUAUUUUUAAUUUA-3′; 2×GAS, 5′-biotin-CGTTTCCCCGAAATTGACGGATTTCCCCGAAAC-3′ (forward) and 5′-GTTTCGGGGAAATCCGTCAATTTCGGGGAAACG-3′ (reverse); nonGAS, 5′-biotin-CGTTTACCCCAAATTGACGGATTTACCCCAAC-3′ (forward) and 5′-GTTGGGGTAAATCCGTCAATTTGGGGTAAACG-3′ (reverse); newMOTIF, 5′-biotin-CGTTTTTGAGGCAATTGACGGATTTTTGAGGCAAC-3′ (forward) and 5′-GTTGCCTCAAAAATCCGTCAATTGCCTCAAAAACG-3′ (reverse). All oligos were purchased from Thermo Fisher Scientific.

ChIP assays

ChIP assay was performed using the Pierce Magnetic ChIP Kit (Thermo Fisher Scientific, 88848) according to the manufacturer’s protocol with minor modifications. Briefly, dTHP-1 cells were treated with or without LPS for 4 hours and then subjected to cross-linking by treatment with 37% formaldehyde (Nacalai Tesque) for 10 min at room temperature. Cross-linking was quenched by treatment with 1.375 M glycine (Nacalai Tesque) for 5 min at room temperature. Nuclei were isolated, and chromatin was sheared by micrococcal nuclease for 5 min at 37°C. Sheared chromatin was harvested by interrupted sonication for 20 cycles at 3 s per cycle. Input sample was prepared with 5% of the sheared chromatin, and the rest of the sample was subjected to immunoprecipitation with anti-STAT1 antibody (1:50 dilution) or IgG according to the manufacturer’s protocol. Both the input and the immunoprecipitates were reverse cross-linked with proteinase K at 65°C overnight. Reverse cross-linked chromatin was eluted and subjected to qRT-PCR analysis. The ChIP-qPCR primers are listed in table S1.

Electrophoretic mobility shift assay

Biotin end-labeled duplex DNA containing two repeats of the 5′-TTTGAGGC-3′ sequence was incubated with cleared extracts from U3A cells transiently overexpressing the appropriate STAT1 constructs. DNA-protein complexes were electrophoresed on a native gel using the LightShift Chemiluminescent EMSA Kit (Thermo Fisher Scientific, 20148) according to the manufacturer’s instructions. DNA-protein complexes were transferred onto Biodyne B Nylon Membrane (Thermo Fisher Scientific) and cross-linked by CL-1000 ultraviolet cross-linker (UVP).

Homology modeling of human and mouse STAT1 proteins

Homology modeling was performed with MODELLER (52). To build a structure of full-length, human STAT1 protein, three PDB templates were used (1bf5_A, 1yvl_A, and 2ka6_B), which cover different regions of the human STAT1 protein sequence. Residue-wise alignment among the PDB templates and the human STAT1 protein sequence was prepared by referring to annotations from the structure integration with function, taxonomy, and sequence (SIFTS) database (53).

Secondary structure prediction of the 3′UTR of IL6 mRNA

The Vienna RNA Package (54) was used to predict the secondary structure of the 3′UTRs of human and mouse IL6 mRNAs. Visualization was performed with Forna (force-directed RNA) (55).

Statistical analysis

Either one-way analysis of variance (ANOVA) with post hoc Tukey’s test or paired two-tailed Student’s t test was used to determine P values. A P value of <0.05 was considered to be statistically significant. All statistical analyses were performed with GraphPad Prism software.

SUPPLEMENTARY MATERIALS

stke.sciencemag.org/cgi/content/full/13/624/eaay0574/DC1

Fig. S1. LPS-induced TLR4 endocytosis promotes ARID5A, IL-6, and IL-12p40 production.

Fig. S2. ARID5A stabilizes IL6 mRNA, but not IL12B or TNF mRNA, through binding to its 3′UTR.

Fig. S3. TLR4 endocytosis–dependent IFN-β–JAK–STAT1-pTyr701 signaling is dispensable for ARID5A expression.

Fig. S4. Phosphorylation of STAT1 at Thr749 facilitates its binding to a noncanonical DNA motif in the ARID5A promoter region.

Fig. S5. A noncanonical TBK1-IKKβ heterodimer mediates the phosphorylation of STAT1 at Thr749 downstream of TLR4 endocytosis.

Table S1. List of primer sequences for RT-qPCR analysis, plasmid construction, and ChIP-qPCR assays.

Table S2. List of antibodies used for Western blotting and immunoprecipitations.

REFERENCES AND NOTES

Acknowledgments: We thank C. Y. Tsai, S. Sakakibara, K. Maeda, H. Kikutani, and S. Akira (IFReC, Osaka University) for their technical help, providing reagents and discussions. We thank T. Akagi (Osaka Bioscience Institute) for providing the pCX4-bsr plasmid. We also thank H. Inoue for technical assistance, and T. Iizuka and M. Okawa for secretarial support. Funding: This work was supported by the Kishimoto Foundation. H.M. was supported by KAKENHI Research Activity Start-up grant (no. 19 K23864). Author contributions: H.M., T.T., and T.K. designed the experiments; H.M. performed the experiments; G.P., S.K., and H.H. helped in the experiments; H.M. and G.P. analyzed the data; S.L. performed bioinformatics analysis; S.K., S.H., J.P.C., and Y.G. provided resources; H.M. wrote the original manuscript; T.T., D.M.S., and T.K. reviewed and edited the manuscript; and T.K. supervised the project. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper or the Supplementary Materials.

Stay Connected to Science Signaling

Navigate This Article